Hi there. Starting on Oct. 22 I started getting dramatically worse response times from Llama 4, to a point where my code would have to time out the request after waiting for up to 3 minutes. This has gotten worse in the past few days.
Is anyone else seeing the same issues?
@shanif can you please provide a specific day and time window for good and bad performance so that we can pull logs . If you have specific request IDs for a good and bad perf example that would be of great use as well.
@omkar.gangan please assit on this one.
-Coby
@shanif can you test again please. There were some queue modifications to favor enterprise accounts . We have made some slight adjustments which should make it bit better for non-enterprise.
-Coby
Will take a look and report back, appreciate the reply
Unfortunately this still seems to be a problem, at least on my end. I just had another request timeout.
Appreciate your assistance with resolving the issue… Sambanova has been my primary LLM provider for the past 6 months and I’ve had to switch to Groq but would love to come back, as I found your models’ answers to be more accurate.
This seems to be resolved now. Thanks for handling!